Create a journey planning agentic workflow with Amazon Nova

Touring is pleasurable, however journey planning may be complicated to navigate and a problem. Vacationers should ebook lodging, plan actions, and prepare native transportation. All these choices can really feel overwhelming. Though journey professionals have lengthy helped handle these complexities, current breakthroughs in generative AI have made one thing totally new potential—clever assistants that may perceive pure dialog, entry real-time information, and immediately interface with reserving methods and journey instruments. Agentic workflows, which use massive language fashions (LLMs) with entry to exterior instruments, are notably promising for simplifying dynamic, multi-step processes like journey planning.

On this put up, we discover the way to construct a journey planning resolution utilizing AI brokers. The agent makes use of Amazon Nova, which provides an optimum stability of efficiency and value in comparison with different business LLMs. By combining correct however cost-efficient Amazon Nova fashions with LangGraph orchestration capabilities, we create a sensible journey assistant that may deal with complicated planning duties whereas maintaining operational prices manageable for manufacturing deployments.

Resolution overview

Our resolution is constructed on a serverless AWS Lambda structure utilizing Docker containers and implements a complete three-layer method: frontend interplay, core processing, and integration providers. Within the core processing layer, we use LangGraph, a stateful orchestration framework, to create a classy but versatile agent-based system that manages the complicated interactions required for journey planning.

The core of our system is a graph structure the place parts (nodes) deal with distinct points of journey planning, with the router node orchestrating the movement of data between them. We use Amazon Nova, a brand new technology of state-of-the-art basis fashions (FMs) obtainable completely on Amazon Bedrock that delivers frontier intelligence with industry-leading price-performance. The router node makes use of an LLM to investigate every consumer question and, with entry to the outline of our 14 motion nodes, decides which of them must be executed. The motion nodes, every with their very own LLM chain, powered by both Amazon Nova Professional or Amazon Nova Lite fashions, handle numerous capabilities, together with internet analysis, customized suggestions, climate lookups, product searches, and procuring cart administration.

We use Amazon Nova Lite for the router and less complicated motion nodes. It may possibly deal with question evaluation and primary content material technology with its lightning-fast processing whereas sustaining robust accuracy at a low value. 5 complicated nodes use Amazon Nova Professional for duties requiring superior instruction following and multi-step operations, corresponding to detailed journey planning and proposals. Each fashions assist a 300,000-token context window and might course of textual content, picture, and video inputs. The fashions assist textual content processing throughout greater than 200 languages, serving to our journey assistant serve a world viewers.The mixing layer unifies a number of information sources and providers by means of an interface:

These integrations function examples, and the structure is designed to be extensible, so organizations can shortly incorporate their very own APIs and information sources based mostly on particular necessities.

The agent retains observe of the dialog state utilizing AgentState (TypedDict), a particular Python dictionary that helps forestall information errors by implementing particular information sorts. It shops the knowledge we have to learn about every consumer’s session: their dialog historical past, profile data, processing standing, and closing outputs. This makes certain the completely different motion nodes can entry and replace data reliably.

The next diagram illustrates the answer structure.

The journey assistant processes consumer interactions from finish to finish:

1. Customers work together with a React.js internet utility by means of a chat interface.
2. Their requests are authenticated utilizing Amazon Cognito and routed by means of Amazon API Gateway.
3. Authenticated requests are despatched to our backend Lambda capabilities, which host the core agent workflow.
4. API credentials are securely saved utilizing AWS Secrets and techniques Supervisor, following finest practices to ensure these delicate keys are by no means uncovered in code or configuration recordsdata, with acceptable entry controls and rotation insurance policies applied.
5. The Journey Assistant Agent itself consists of a number of interconnected parts. On the middle, the agent router analyzes incoming queries and orchestrates the workflow.
6. The agent maintains state by means of three DynamoDB tables that retailer dialog historical past, procuring wishlists, and consumer profiles, ensuring context is preserved throughout interactions.
7. For travel-specific information, the system makes use of a mix of Amazon Bedrock Data Bases, Amazon OpenSearch Serverless, and a doc retailer in Amazon Easy Storage Service (Amazon S3). These parts work collectively to offer correct, related journey data when wanted.
8. The agent’s motion nodes deal with specialised duties by combining LLM chains with exterior APIs. When customers want product suggestions, the system connects to the Amazon Product Promoting API. For common journey data, it makes use of the Google Customized Search API, and for weather-related queries, it consults the OpenWeather API. API credentials are securely managed by means of Secrets and techniques Supervisor.
9. The system formulates complete responses based mostly on collected data, and the ultimate responses are returned to the consumer by means of the chat interface.

This structure helps each easy queries that may be dealt with by a single node and complicated multi-step interactions that require coordination throughout a number of parts. The system can scale horizontally, and new capabilities may be added by introducing further motion nodes and API integrations.

You may deploy this resolution utilizing the AWS Cloud Improvement Equipment (AWS CDK), which generates an AWS CloudFormation template that handles the required assets, together with Lambda capabilities, DynamoDB tables, and API configurations. The deployment creates the required AWS assets and outputs the API endpoint URL on your frontend utility.

Conditions

For this walkthrough, you have to have the next stipulations:

Clone the repository

Begin by cloning the GitHub repository containing the answer recordsdata:

git clone https://github.com/aws-samples/sample-travel-assistant-agent.git

Acquire API keys

The answer requires API keys from three providers to allow its core functionalities:

• OpenWeather API – Create a Free Entry account at OpenWeather to acquire your API key. The free tier (60 calls per minute) is adequate for testing and improvement.
• Google Customized Search API – Arrange the search performance by means of Google Cloud Console. Create or choose a challenge and allow the Customized Search API. Then, generate an API key from the credentials part. Create a search engine at Programmable Search and observe your Search Engine ID. The free tier contains 100 queries per day.
• (Optionally available) Amazon Product Promoting API (PAAPI) – If you wish to allow product suggestions, entry the PAAPI Documentation Portal to generate your API keys. You’ll obtain each a public key and a secret key. You need to have an Amazon Associates account to entry these credentials. In the event you’re new to the Amazon Associates Program, full the appliance course of first. Skip this step should you don’t wish to use PAAPI options.

Add API keys to Secrets and techniques Supervisor

Earlier than deploying the answer, you have to securely retailer your API keys in Secrets and techniques Supervisor. The next desk lists the secrets and techniques to create and their JSON construction. For directions to create a secret, check with Create an AWS Secrets and techniques Supervisor secret.

Configure surroundings variables

Create a .env file within the challenge root along with your configuration:

STACK_NAME=TravelAssistantAgent

# Optionally available: Create Bedrock Data Base with paperwork
KB_DOCS_PATH = Path/to/your/paperwork/folder
# Optionally available: Allow/disable Product Search options with PAAPI
USE_PAAPI=false

Deploy the stack

If that is your first time utilizing the AWS CDK in your AWS account and AWS Area, bootstrap your surroundings:

Deploy the answer utilizing the offered script, which creates the required AWS assets, together with Lambda capabilities, DynamoDB tables, and API configurations:

Entry your utility

When the deployment is full, open the AWS CloudFormation console and open your stack. On the Outputs tab, observe the next values:

• WebAppDomain – Your utility’s URL
• UserPoolId – Required for consumer administration
• UserPoolClientId – Used for authentication

Create an Amazon Cognito consumer

Full the next steps to create an Amazon Cognito consumer:

1. On the Amazon Cognito console, select Person swimming pools within the navigation pane.
2. Select your consumer pool.
3. Select Customers within the navigation pane, then select Create consumer.

4. For Electronic mail deal with, enter an e mail deal with, and choose Mark e mail deal with as verified.
5. For Password, enter a brief password.
6. Select Create consumer.

You should utilize these credentials to entry your utility on the WebAppDomain URL.

Take a look at the answer

To check the agent’s capabilities, we created a enterprise traveler persona and simulated a typical journey planning dialog movement. We targeted on routing, operate calling accuracy, response high quality, and latency metrics. The agent’s routing system directs the consumer inquiries to the suitable specialised node (for instance, trying to find lodging, checking climate circumstances, or suggesting journey merchandise). All through the dialog, the agent maintains the context of beforehand mentioned particulars, so it will probably construct upon earlier responses whereas offering related new data. For instance, after discussing journey vacation spot, the agent can naturally incorporate this into subsequent climate and packing record suggestions.

The next screenshots show the end-user expertise, whereas the underlying API interactions are dealt with seamlessly on the backend. The whole implementation particulars, together with Lambda operate code and API integration patterns, can be found in our GitHub repository.

The answer demonstrates personalization capabilities utilizing pattern consumer profiles saved in DynamoDB, containing upcoming journeys and journey preferences. In manufacturing deployments, these profiles may be built-in with current buyer databases and reservation methods to offer a customized help.

The product suggestions proven are dwell hyperlinks to precise gadgets obtainable on Amazon.com, so the consumer can discover or buy these merchandise immediately. The consumer can select a hyperlink to take a look at the product, or select Add to Amazon Cart to see the gadgets of their procuring cart.

Clear up

After you’re completed experimenting with the journey assistant, you’ll be able to find the CloudFormation stack on the AWS CloudFormation console and delete it. It will delete the assets you created.

Conclusion

Our journey planning assistant agent demonstrates a sensible utility constructed by Amazon Nova and LangGraph for fixing real-world enterprise challenges. The system streamlines complicated journey planning whereas naturally integrating product suggestions by means of specialised processing nodes and real-time information integration. Amazon Nova Lite fashions confirmed cheap efficiency at activity orchestration, and Amazon Nova Professional carried out effectively for extra complicated operate calling operations. Wanting forward, this framework could possibly be applied with extra dynamic orchestration methods corresponding to ReAct. To construct your individual implementation, discover our code samples within the GitHub repository.

For these seeking to deepen their understanding of LLM-powered brokers, AWS supplies in depth assets on constructing clever methods. The Amazon Bedrock Brokers documentation provides insights into automating multistep duties with FMs, and the AWS Bedrock Agent Samples GitHub repo supplies steering for implementing a number of agent purposes utilizing Amazon Bedrock.

In regards to the authors

Isaac Privitera is a Principal Information Scientist with the AWS Generative AI Innovation Heart, the place he develops bespoke generative AI-based options to deal with clients’ enterprise issues. His main focus lies in constructing accountable AI methods, utilizing methods corresponding to RAG, multi-agent methods, and mannequin fine-tuning. When not immersed on this planet of AI, Isaac may be discovered on the golf course, having fun with a soccer sport, or mountaineering trails along with his loyal canine companion, Barry.

Ryan Razkenari is a Deep Studying Architect on the AWS Generative AI Innovation Heart, the place he makes use of his experience to create cutting-edge AI options. With a powerful background in AI and analytics, he’s keen about constructing revolutionary applied sciences that deal with real-world challenges for AWS clients.

Sungmin Hong is a Senior Utilized Scientist on the AWS Generative AI Innovation Heart, the place he helps expedite quite a lot of use instances for AWS clients. Earlier than becoming a member of Amazon, Sungmin was a postdoctoral analysis fellow at Harvard Medical College. He holds a PhD in Laptop Science from New York College. Outdoors of labor, Sungmin enjoys mountaineering, studying, and cooking.